In short
- Definition: The ability of a system to keep operating correctly when one or more of its components fail.
- Goal: Prevent single failures from interrupting the service.
- Example S.M.A.R.T. requirement: No user-visible outage when any single server or zone is shut down in monthly chaos tests.
1. Definition
As a non-functional requirement, fault tolerance describes the ability of a system to keep operating correctly when one or more of its components fail: a server, a disk, a network link, a data center zone or a dependency [1], [2].
A single point of failure is a requirement nobody wrote down, but everybody will discover.Key takeaway
In ISO/IEC 25010:2023, fault tolerance is a sub-characteristic of reliability. It is achieved through redundancy (several copies of each component), automatic failover, isolation between parts and graceful handling of errors [3].
Fault tolerance is tested by deliberately causing failures. Chaos engineering describes this as experimenting on a system to build confidence in its ability to withstand turbulent conditions in production [4].
2. Goal
The goal is to prevent a single failure from interrupting the service for users.
Without fault tolerance, any one broken component, including ones that fail routinely such as disks and servers, becomes an outage.
3. How to Measure It (S.M.A.R.T.)
| Metric | Unit | Typical target | How to measure |
|---|---|---|---|
| User-visible outage during single-component failure | Minutes | 0 | Failure injection tests |
| Single points of failure | Count | 0 on critical paths | Architecture review |
| Failover time | Seconds | < 30 s | Measured during failover tests |
| Chaos experiments | Per month | ≥ 1 on production-like environment | Experiment log |
4. Example Requirements
Each requirement below is ready to copy into a specification: specific, measurable, achievable, relevant and time-bound.
- Web service: “Shutting down any single server or availability zone during the monthly chaos test shall cause no user-visible outage and less than 0.1% failed requests.”
- Database: “The database shall fail over automatically to a replica within 30 seconds, without losing committed transactions.”
- Dependencies: “If the recommendation service fails, product pages shall still load, without recommendations.”
5. Questions to Ask Stakeholders
- Which components could fail, and what happens to users when they do?
- Are there single points of failure today?
- Which failures must be invisible to users, and which can degrade the service?
- How fast must failover happen?
- Can we test failures in production or a realistic copy?
- What is the budget for redundancy?
6. Common Pitfalls
- Redundancy that is never tested. Failover paths rot. Exercise them regularly.
- Shared hidden dependencies. Two replicas relying on the same DNS, network or configuration can fail together.
- Retry storms. Unlimited retries during a failure can overload the remaining components. Use backoff and limits.
7. Related NFRs
- Availability — fault tolerance keeps availability high when parts fail.
- Reliability — ISO/IEC 25010 places fault tolerance under reliability.
- Resilience — covers degraded operation and recovery beyond simple failover.
- Robustness — handles bad inputs and unexpected conditions.
- Failure Management — detects and handles failures that are not absorbed.
8. FAQ
Fault tolerance aims for no interruption when a component fails; high availability accepts short interruptions but keeps total downtime low.
The practice of running controlled experiments, such as killing servers, to check that the system withstands failures.
Not always. Focus redundancy on critical paths and on components that fail most often.
9. Acronyms, Initialisms & Terms
Acronyms are abbreviations pronounced as words (like NASA); initialisms are spelled out letter by letter (like FBI). The classification below follows common pronunciation.
Acronyms
| Abbreviation | Stands for | Definition | Ref. |
|---|---|---|---|
| ISO | International Organization for Standardization | Independent international body that publishes standards, such as ISO/IEC 25010 for software quality. Its short name comes from the Greek “isos” (equal) and is said as a word. | [5] |
| SMART | Specific, Measurable, Achievable, Relevant, Time-bound | Criteria for writing a goal or requirement so that anyone can check whether it has been met. | [6] |
Initialisms
| Abbreviation | Stands for | Definition | Ref. |
|---|---|---|---|
| DNS | Domain Name System | The internet’s directory: it translates names such as example.com into network addresses. | [7] |
| FAQ | Frequently asked questions | List of common questions with their answers. | [8] |
| IEC | International Electrotechnical Commission | International standards body for electrical and electronic technologies; it publishes joint “ISO/IEC” standards with ISO. | [9] |
| NFR | Non-functional requirement | Requirement that describes how well a system must work (its qualities, such as speed or security) rather than what it does. | [2] |
Terms
Plain-language definitions of technical and project-management terms used in this article.
| Term | Plain-language definition | Ref. |
|---|---|---|
| Chaos engineering | Deliberately injecting failures in a controlled way to check that a system copes. | [10] |
| Data center | Building that houses servers, storage and network equipment. | [11] |
| Database | Organized store of data that software can query and update. | [12] |
| Failover | Automatic switch to a standby component when the main one fails. | [13] |
| Log | Time-stamped record of events written by software, used for troubleshooting and audit. | [14] |
| Production environment | The live system that real users use, as opposed to test environments. | [15] |
| Redundancy | Duplicating critical components so that one failure does not stop the service. | [16] |
| Replication | Keeping synchronized copies of data or services on several machines. | [17] |
| Stakeholder | Person or group with an interest in the system: users, sponsors, operators, regulators. | [18] |
| Transaction | Unit of work, such as a payment or an order, that must complete fully or not at all. | [19] |
10. Sources and Further Reading
- [1] “Fault tolerance.” Wikipedia. https://en.wikipedia.org/wiki/Fault_tolerance (accessed Nov. 18, 2026).
- [2] “Non-functional requirement.” Wikipedia. https://en.wikipedia.org/wiki/Non-functional_requirement (accessed Nov. 18, 2026).
- [3] “ISO/IEC 25010 product quality model.” ISO 25000. https://iso25000.com/index.php/en/iso-25000-standards/iso-25010 (accessed Nov. 18, 2026).
- [4] principlesofchaos.org. “Principles of Chaos Engineering.” Principles of Chaos Engineering. https://principlesofchaos.org/ (accessed Nov. 18, 2026).
- [5] “International Organization for Standardization.” Wikipedia. https://en.wikipedia.org/wiki/International_Organization_for_Standardization (accessed Nov. 19, 2026).
- [6] “SMART criteria.” Wikipedia. https://en.wikipedia.org/wiki/SMART_criteria (accessed Nov. 19, 2026).
- [7] “Domain Name System.” Wikipedia. https://en.wikipedia.org/wiki/Domain_Name_System (accessed Nov. 19, 2026).
- [8] “FAQ.” Wikipedia. https://en.wikipedia.org/wiki/FAQ (accessed Nov. 19, 2026).
- [9] “International Electrotechnical Commission.” Wikipedia. https://en.wikipedia.org/wiki/International_Electrotechnical_Commission (accessed Nov. 19, 2026).
- [10] “Chaos engineering.” Wikipedia. https://en.wikipedia.org/wiki/Chaos_engineering (accessed Nov. 19, 2026).
- [11] “Data center.” Wikipedia. https://en.wikipedia.org/wiki/Data_center (accessed Nov. 19, 2026).
- [12] “Database.” Wikipedia. https://en.wikipedia.org/wiki/Database (accessed Nov. 19, 2026).
- [13] “Failover.” Wikipedia. https://en.wikipedia.org/wiki/Failover (accessed Nov. 19, 2026).
- [14] “Logging (computing).” Wikipedia. https://en.wikipedia.org/wiki/Logging_(computing) (accessed Nov. 19, 2026).
- [15] “Deployment environment.” Wikipedia. https://en.wikipedia.org/wiki/Deployment_environment (accessed Nov. 19, 2026).
- [16] “Redundancy (engineering).” Wikipedia. https://en.wikipedia.org/wiki/Redundancy_(engineering) (accessed Nov. 19, 2026).
- [17] “Replication (computing).” Wikipedia. https://en.wikipedia.org/wiki/Replication_(computing) (accessed Nov. 19, 2026).
- [18] “Project stakeholder.” Wikipedia. https://en.wikipedia.org/wiki/Project_stakeholder (accessed Nov. 19, 2026).
- [19] “Database transaction.” Wikipedia. https://en.wikipedia.org/wiki/Database_transaction (accessed Nov. 19, 2026).
Written by Philippe Huard · LinkedIn · Last updated: November 18, 2026 · Header photo by @lukechesser; inline photo by @pathdigital, both on Unsplash.




